def search_datasets(tags=None, version=None, ontologies=None):
"""
:param tags: List of strings
:param version: List of strings
:ontologies: List of ontology terms
:return: List of datasets matching any of the supplied parameters
"""
db_session = get_session()
print(tags, version, ontologies)
try:
datasets = db_session.query(Dataset)
if version:
datasets = datasets.filter(
Dataset.version.like('%' + version + '%'))
if tags:
# return any project that matches at least one tag
datasets = datasets.filter(
or_(*[Dataset.tags.contains(tag) for tag in tags]))
if ontologies:
datasets = datasets.filter(
or_(*[
Dataset.ontologies_internal.contains(term)
for term in ontologies
]))
except ORMException as e:
err = _report_search_failed('dataset', e)
return err, 500
return [dump(x) for x in datasets], 200
def delete_dataset_by_id(dataset_id):
"""
Current thoughts are that delete should only be a CLI accessible command
rather than API
:param dataset_id: UUID
:return: 204 on successful delete
"""
db_session = get_session()
try:
specified_dataset = db_session.query(Dataset) \
.get(dataset_id)
except ORMException as e:
err = _report_search_failed('call', e, dataset_id=str(dataset_id))
return err, 500
if not specified_dataset:
err = dict(message="Dataset not found: " + str(dataset_id), code=404)
return err, 404
try:
row = db_session.query(Dataset).filter(
Dataset.id == dataset_id).first()
db_session.delete(row)
db_session.commit()
except ORMException as e:
err = _report_update_failed('dataset', e, dataset_id=str(dataset_id))
return err, 500
return None, 204
def get_versions():
"""
:return: release versions of the database
"""
db_session = get_session()
change_log = ChangeLog
try:
versions = db_session.query(change_log.version)
except ORMException as e:
err = _report_search_failed('versions', e)
return err, 500
return [entry.version for entry in versions], 200
def get_change_log(version):
"""
:param version: required release version
:return: changes associated with specified release version
"""
db_session = get_session()
change_log = ChangeLog
try:
log = db_session.query(change_log)\
.get(version)
except ORMException as e:
err = _report_search_failed('change log', e)
return err, 500
if not log:
err = dict(message="Change log not found", code=404)
return err, 404
return dump(log), 200
def get_dataset_by_id(dataset_id):
"""
:param dataset_id: UUID
:return: all projects or if projectId specified, corresponding project
"""
db_session = get_session()
try:
validate_uuid_string('id', dataset_id)
specified_dataset = db_session.query(Dataset) \
.get(dataset_id)
except IdentifierFormatError as e:
err = dict(message=str(e), code=404)
return err, 404
if not specified_dataset:
err = dict(message="Dataset not found: " + str(dataset_id), code=404)
return err, 404
return dump(specified_dataset), 200
def post_change_log(body):
"""
Create a new change log following the changeLog
schema in datasets.yaml
:return: body, 200 on success
"""
db_session = get_session()
change_version = body.get('version')
body['created'] = datetime.datetime.utcnow()
try:
orm_changelog = ChangeLog(**body)
except TypeError as e:
err = _report_conversion_error('changelog', e, **body)
return err, 400
try:
db_session.add(orm_changelog)
db_session.commit()
except exc.IntegrityError:
db_session.rollback()
err = _report_object_exists('changelog: ' + body['version'], **body)
return err, 405
except ORMException as e:
err = _report_write_error('changelog', e, **body)
return err, 500
logger().info(
struct_log(action='post_change_log',
status='created',
change_version=change_version,
**body))
return body, 201
def search_dataset_ontologies():
"""
Return all ontologies currently used by datasets
"""
db_session = get_session()
try:
datasets = db_session.query(Dataset)
valid = datasets.filter(Dataset.ontologies != [])
ontologies = [dump(x)['ontologies'] for x in valid]
terms = sorted(
list(
set([
term['id'] for ontology in ontologies for term in ontology
])))
except ORMException as e:
err = _report_search_failed('dataset', e)
return err, 500
return terms, 200
def post_dataset(body):
"""
Creates a new dataset following the dataset_ingest
schema defined in datasets.yaml
The ontologies_internal property is used when looking
up current ontologies but is not a property to be returned
when querying the dataset.
"""
db_session = get_session()
if not body.get('id'):
iid = uuid.uuid1()
body['id'] = iid
else:
iid = body['id']
if not body.get('version'):
body['version'] = Version
body['created'] = datetime.datetime.utcnow()
mapped = []
if body.get('ontologies'):
# Ontology objects should be {'id': ontology_name, 'terms': [{'id': 'some code'}]}
mapped = {
ontology['id']: ontology['terms']
for ontology in body['ontologies']
}
if 'duo' in mapped.keys():
validator = OntologyValidator(ont=ont, input_json=mapped)
valid, invalids = validator.validate_duo()
if not valid:
err = dict(message="DUO Validation Errors encountered: " +
str(invalids),
code=400)
return err, 400
duo_terms = json.loads(validator.get_duo_list())
duos = []
for term in duo_terms:
stuff = OntologyParser(ont, term["id"]).get_overview()
duos.append({**term, **stuff})
body['ontologies'] = duos
body['ontologies_internal'] = mapped
try:
orm_dataset = Dataset(**body)
except TypeError as e:
err = _report_conversion_error('dataset', e, **body)
return err, 400
try:
db_session.add(orm_dataset)
db_session.commit()
except exc.IntegrityError:
db_session.rollback()
err = _report_object_exists('dataset: ' + body['id'], **body)
return err, 405
except ORMException as e:
db_session.rollback()
err = _report_write_error('dataset', e, **body)
return err, 500
body.pop('ontologies_internal')
return body, 201